As a marketer, I’m a fan of open source software. After all, if you can’t dislodge Microsoft from mid-range server operating systems, Microsoft Office from desktop productivity suites, or Oracle from relational databases — and doing so through traditional means is a virtual impossibility — then blowing up the whole business model isn’t a bad start. It’s creative and it cuts right to the core of the problem.
But as a business-person I am not. When you play the role of market spoiler it’s much easier to be famous than rich. For example, when MySQL was acquired by Sun in 2008 for $1.2B, MySQL was doing only about $65M in annual revenues. While the revenue multiple on the exit was spectacular, their capture rate was not: MySQL disrupted literally billions in “big three” (i.e., Oracle, DB2, SQL Server) database revenues. But if your value proposition is rooted in “almost free relative to leading commercial alternatives,” then you won’t succeed at 50% of their cost; you’ll need to be more like 2-5%.
I refer to open source as both a development model — i.e., a way of building software — and a business model. While the former is more well defined than the latter, the typical way to make money in open source is through selling subscriptions or licenses to certified and more-quickly-patched releases as well as selling technical support and/or consulting services to go with them.
While a spectacular exit multiple may occasionally pay off big time for shareholders (e.g., JBoss, MySQL), my theory is that in general it’s very hard to make money with the open source business model. Red Hat is the obvious exception, and we’ll talk about them in a minute.
The basic paradox of open source is this:
- The smaller the community the worse the software quality and the more people need certified releases and support.
- The bigger the community the higher quality the software and the less people need certified releases and support (i.e., the community version will do).
So you can have a large community who doesn’t need to buy from you or a small community who does.
Two other drivers complete the picture:
- The nature of the software and to what extent it truly requires an almost-daily stream of patches and updates and …
- The monetization rate which is a function of the commercial market structure. For example, the lower-level the software (e.g., operating systems) the more the market tends towards natural monopoly as customers want to minimize entropy at the bottom of the stack. This should drive high pricing/margins on the commercial side of the market, and a parallel opportunity for someone to establish clear leadership on the open source side.
This is why Red Hat does so well when most others end up stagnating in the tens-of-millions of revenues range. The market is huge. The software is low-level and thus the market “wants” a clear leader (think: increasing returns) who can provide a hardware-independent, low-cost, supported product as an alternative to the proprietary Unix-es of days past.
Put differently, the bigger the commercial market and the more monopolistic its structure, the better the open source opportunity. Conversely, the smaller the commercial market and the more fragmented leadership is within it (e.g., enterprise search, document management, and to some extent BI), the worse the open source opportunity.
Well said, Dave (as usual). From my experience, most open-source business models end up as lifestyle business. Get to the tens of millions in revenue, keep your costs low and don’t take any outside investor money — and you might end up with a small, but relatively profitable, business.
It also worries me that so many large open source projects are funded by companies who get that funding from some other revenue stream, meaning that the open source project is also a cost centre, never a profit centre. That is never going to be sustainable in the longer term.
I think RedHat is a special case. RedHat makes its money, in my opinion, because you have to pay for the version of the software that is approved for use with big closed source applications like Oracle. If people were only running open source applications on RedHat, I think they would be far less inclined to pay as much for it.
I heard an interesting purchasing guideline from friends at BT, along the line of “use open source for horizontal applications, vendor solutions for vertical applications, and your own code for niche applications”. That’s not bad advice for buyers, but if vendors are forced into the vertical markets, I wonder how much money will be left to support the horizontal applications.
Thanks Jim. I agree. If you properly assess the market size, you can run a small profitable business in many of these categories.
Great point Tony. Several open source projects are partially funded / supported by companies hoping to undermine the monopolist/leader’s position in strategies presumably designed to weaken one’s opponent by trying to damage their core business by taking funds out of yours.
This suggests that should they ever succeed, their commitment might suddenly drop. Basically, the projects are funded as spoilers more than opportunities unto themselves.
BT’s advice makes some sense to me, though I think it misses the opportunity to use open source winners at low levels in the stack. But those aside, what I think they’re saying is go get tactical wins using open source but don’t bet big strategically (think: use broadly and/or as a platform) unless it’s already a winner.
Very well written Dave. As you put it open source model is a big disruptor to the market. The ones that are successful put functionality and community over marketing . Redhat is a very special case that has done well in all aspect and even in marketing. One comment that you have made that I agree with you is; that for open source software to survive “The Market demand for that type of software has to be huge”
Good Article, John
Pingback: links for 2011-05-10 « steinarcarlsen
Interesting post Dave. I agree entirely that you need both a large market and a market which tends towards consolidation behind a leader. Red Hat, JBoss, Spring, and MySQL all had those.
My view is that Red Hat, JBoss, and SpringSource were all successful, capturing 5-10% of the value from an entrenched leader in a settled market; I’d benchmark RedHat to Microsoft’s OS business and JBoss/SpringSource against BEA.
MySQL didn’t do nearly as well, with closer to 1% of Oracle’s database business. I’d argue the difference was product differentiation. There were plenty of people who thought Linux was way better than Windows NT (which I think was the more relevant competitor than Solaris); it certainly was differentiated for better or worse. On the other hand you won’t find many who think MySQL is a better database than Oracle. As a result, I believe the MySQL user base had much stronger selection bias towards price resistance than did Red Hat’s.
So I’d add a third requirement for success – product differentiation – and while MySQL was a success from an investor perspective, I’d argue that it failed to capture the “normal” share of value for a broadly adopted technology in an infrastructure technology.
Interesting point, Max. You’re arguing that if you’re not only cheaper — but better — then you can have a 5-10% capture rate as opposed to a 1-5% rate. Makes sense to me on first blush. I’ll think about it more and get comment again if I have additional insights.
Dave – not sure I agree 100% with your assertion that crappy software = more monetization, and good software = less monetization
You have to lead with a product that has certain tangible benefits that compel people to use your product.
For RedHat it is linux standardization, and for MongoDB (I work at 10gen), there are other differentiators (e.g. ease of development, scalability, etc). Your licensing/distribution/monetization strategy then needs to support this benefit. RedHat is doing well because their business model (paid binaries + certification + support) matches well with their benefit (standardization + low entropy at the OS layer)
As a side note, AFAIK there were markets that MySQL monetized fairly well e.g. the OEM market. There the benefit was an easy to imbed, but fully-featured relational database and the business model (OEM licenses) matched well.
Anyway, a very good and thought provoking post!
If you are interested, I wrote up a post a few weeks ago that talked about some of the other (non-monetization related) benefits of an open source business model:
Thanks Nosh. I think the inverse relationship between community size and software quality should generally hold (and I heard from a friend who’s been at a moderately popular open source company for many years). But I would be hard-pressed to prove it.
I think the most interesting point that you and a few others made to me over email was what about markets which are asymmetric. I was thinking about spaces where I personally know lots of people (and hear lots of stories), e.g., enterprise search, database, document/content management. In these spaces, the markets are symmetric (i.e., you can buy Autonomy IDOL search or use Lucene, or you can buy Oracle or use MySQL or you can buy Documentum or use Al Fresco) — i.e., you can use enterprise thing X or its open source equivalent. That’s primarily what I was thinking about when I wrote the post, including I should add — BI.
I had some feedback from the NoSQL crowd reminding me of the asymmetry of those markets. There is no classic, enterprise commercial equivalent of MongoDB, for example. In these markets, then product differentiation comes strongly into play.
On MySQL, I agree with your point that the business model matched some markets better than others as per your example. I’ll check out your post as well. Best/Dave
Pingback: Open Source Business Models, Revisited | Kellblog